研究背景和目的：按疾病诊断相关分组（Diagnosis Related Group，DRG）在2021年被国家医保局正式向全国推广，将在2024年成为中国医保主要的付费方式。在控制医疗费用增长和促进医疗服务规范性提升的同时，也给医保管理、医疗机构绩效考核等工作带来不确定性，并可能对医疗行为产生影响。重症医学科（Intensive Care Unit，ICU）收治的患者病情复杂，治疗费用高昂，医疗行为直接关系患者的生命存亡。目前DRG付费实施对ICU医疗行为影响的实证研究还比较少。本研究基于激励相容理论，探讨DRG付费实施对ICU医疗行为的影响，分析医保管理者、医院管理者和ICU医务人员等对DRG付费实施的看法，探索DRG付费实施影响ICU医疗行为的可能机制，提出针对性的政策建议。
研究方法：本研究采用定量研究和定性研究相结合的混合研究方法。首先，本研究选取天津市2018年1月-2022年9月期间的12家三级医院，其中7家在2021年10月实施DRG付费，作为干预组；其余5家尚未实施DRG付费的医院作为对照组。研究对象为ICU收治的住院患者（入院科室为ICU或由某住院科室转入），利用电子病例和医院信息系统数据，采用分段线性回归分析，分析DRG付费实施对三类指标（转入ICU患者数、气管切开使用和白蛋白使用）的影响。进而，选取与DRG付费实施关系密切的利益相关者（医保管理人员、医疗机构高层和中级管理者、ICU医务人员、其他科室医务人员）22名，进行半结构式访谈，通过类属分析法探讨DRG付费实施对ICU医疗行为的影响，以及其对DRG付费实施的看法及完善建议，分析DRG付费实施影响ICU医疗行为的可能机制。
研究结果：
1. 定量研究发现，DRG付费实施后：干预组医疗机构其他科室向ICU转诊的患者人数下降，但是与对照组相比，其趋势变化无统计学显著性差异；在ICU气管切开使用上，干预组整体上呈下降趋势，与对照组相比存在统计学显著性差异（P＜0.05）；在ICU白蛋白使用上，干预组和对照组相比趋势变化无统计学差异。
2. 质性研究发现：（1）DRG付费实施后，为了避免科室发生亏损以及影响医务人员个人绩效，某些科室向ICU的转诊意愿下降，表现为转诊患者数量减少以及延误患者转诊，部分ICU医务人员为了获得更好的医保补偿会增加气管切开的使用，但是这些行为的出现不具有普遍性，其出现原因是部分医疗机构绩效考核制度与DRG付费政策不匹配，出现激励不相容。（2）两个层面的制度设计会影响DRG付费实施对ICU产生的影响，第一个层面是医保中心对医疗机构结算时采用的付费制度设计，包括医保费用支出结构、权重设定及协商机制等，第二个层面是医疗机构内部制度设计，特别是医务人员绩效考核制度，优化这两个层面的制度设计可望消除或减少DRG付费实施对ICU医疗行为产生的影响。
结论与讨论：DRG付费实施对ICU部分医疗行为如气管切开的使用有显著影响，但是对ICU转入患者人数和白蛋白使用无显著影响，可能原因：（1）DRG付费实施导致ICU临床医师为控制医疗成本而倾向于采取保守治疗，从而减少气管切开的使用；（2）ICU作为平台科室患者来自多个临床科室，如果某个科室转诊人数下降而其他科室增加或不变，可能导致向ICU转诊总人数不发生变化；干预组包括7家医疗机构、对照组包括5家医疗机构，医疗机构之间如果转诊人数变化不一致，也会导致总体转诊人数不发生变化。（3）影响白蛋白使用的主要因素是临床需求和医保报销标准。其他科室向ICU的转诊行为整体上未出现变化，但是在某些医疗机构会出现其他科室向ICU转诊意愿下降的现象，表现为转诊数量减少和转诊延误，直接原因可能是绩效与科室结余挂钩的政策使医务人员担心转诊ICU可能导致科室亏损及影响个人绩效，深层次原因是绩效与科室结余挂钩的政策与DRG政策目标不匹配，没有形成激励相容。建议未来重点关注医保中心付费、医疗机构绩效考核两个层面的制度设计，使医保中心优化预算管理和调控职能、使医疗机构优化对医务人员的绩效考核办法，结合细化DRG分组和特病单议等手段，实现付费制度与DRG付费政策的激励相容，正向激励医疗机构和医务人员提供急危重症治疗，消除DRG付费实施对ICU带来的影响，使ICU医务人员能够专心于临床诊疗，保障患者生命安全。
 

Background and purposes: With the continuous development of China's healthcare security, the healthcare security system is also facing new challenges. The financial expenditure on healthcare continues to increase, and the income of the medical insurance fund is difficult to continue to increase. As one of the widely adopted  payment methods in the world, Diagnosis Related Group (DRG) has been officially promoted nationwide by the National Healthcare Security Administration in 2021 after years of practice and pilot in China, and will become the main payment method in China by 2024. Existing studies have shown that DRG payment, as a pre-payment system, has been adopted by many countries around the world, which can guide medical institutions to reduce unreasonable expenses and improve medical efficiency. However, it will also affect the medical behaviors of medical institutions, resulting in a series of unexpected behaviors such as patient selection, shifting of severe diseases and service reduction. Intensive Care Unit (ICU) treatment is expensive, but the patients admitted have complex conditions, and if there is unexpected behavior, the patient's life safety will be seriously endangered. However, there are still few studies on the impact of DRG implementation on ICU medical behavior, and empirical studies are scarce. Therefore, based on incentive compatibility theory, this study explores to understand the impact of DRG payment on ICU medical behavior, and analyzes the views of healthcare security administrators, hospital administrators and ICU medical staff on DRG payment implementation, with the aim to explore the possible mechanism of DRG payment’s impact on ICU medical behavior, and put forward targeted policy recommendations.

Methods: This study adopts a mixed research method which combines quantitative research and qualitative research. Firstly, this study selected 12 tertiary hospitals in Tianjin from January 2018 to September 2022, among which 7 hospitals implementing DRG payment in October 2021 as the intervention group; The remaining 5 were the control group and had not implemented DRG payment at the end of the study period. The study subjects were inpatients admitted to ICU (admitted to ICU directly or transferred to ICU from an inpatient department). The original records from two hospital information systems, including electronic medical records and hospital information system, were used to analyze the impact of DRG payment on the three indicators by piecewise-linear regression analysis. Then, 22 stakeholders (medical insurance administrators, senior and middle managers of medical institutions, medical staff in ICU and medical staff in other departments) who are closely related to the implementation of DRG payment were selected and semi-structured interviews were conducted to explore the impact of DRG payment implementation on medical behavior in ICU through the category analysis method. This paper also analyzes the possible mechanism of DRG payment affecting ICU medical behavior.

Results: 1. Quantitative study found that after the implementation of DRG payment, the number of patients referred to ICU from other departments of medical institutions in the intervention group decreased, but there was no statistical difference in the trend compared with the control group; The use of tracheotomy in ICU showed a downward trend in the intervention group as a whole, and there was a statistically significant difference compared with the control group (P < 0.05). There was no significant difference in the trend of albumin use in ICU between the intervention group and the control group. 2. The qualitative study found that (1) After the implementation of DRG payment, in order to avoid department losses and loss of personal performance of medical staff, the willingness of other departments in medical institutions to refer patients to ICU decreased, which was manifested as a decrease in the number of referred patients and an increase in the severity of patients' diseases caused by delayed patient referral. There has also been an increase in the use of tracheostomies by ICU medical staff in order to receive better compensation, but these behaviors are not seen in all medical facilities and departments. (2) The system design at two levels will affect the impact of DRG payment implementation on ICU. The first level is the payment system design adopted by the medical insurance center in the settlement of medical institutions, including the expenditure structure, weight and negotiation mechanism of medical insurance expenses; the second level is the internal system design of medical institutions, especially the performance appraisal system of medical personnel. Optimizing the two levels of institutional design can reduce or eliminate the impact of DRG payment implementation on ICU medical behavior.

Conclusions: DRG payment has an effect on the use of some medical behaviors in ICU, such as tracheotomy, but has no effect on the number of patients transferred to ICU and the use of albumin. Possible reasons are as follows: (1) After the conversion of payment methods, ICU physicians are inclined to adapted conservative treatment measures and reduce the use of tracheostomies to manage clinical expenses. (2) As a platform department, ICU patients come from multiple clinical departments. If the number of referrals in one department decreases while the number of other departments increases or remains unchanged, the total number of referrals to ICU may not change; The intervention group included 7 medical institutions and the control group included 5 medical institutions. If the number of referrals changed inconsistently among medical institutions, the overall number of referrals would not change. (3) The main factors influencing the use of albumin are clinical demand and medical insurance policies. There was no change in the referral behavior of other departments to ICU on the whole, but in some medical institutions, there was a decline in the willingness of other departments to refer patients to ICU, which was manifested as a decrease in the number of referrals and a delay in referral. The reason may be that the policy of linking performance with department balance made medical staff worry that referral to ICU might lead to department losses and affect individual performance i.e. incentive compatibility is not generated. It is suggested that in the future, to generate incentive compatibility, we should focus on the system design at the two levels of payment of medical insurance center and performance appraisal of medical institutions, so that the medical insurance center can optimize its budget management and regulation functions, and medical institutions can optimize the performance appraisal methods for medical personnel. In combination with the means of refining DRG grouping and special disease list, medical institutions and medical personnel can be positively encouraged to provide emergency and critical treatment. It can eliminate the impact and influence of DRG payment implementation on ICU, so that ICU medical staff can concentrate on clinical diagnosis and treatment, and ensure the life safety of patients.